基于 Bi-LSTM 和 Kalman 的光伏发电功率超短期预测

打开文本图片集
关键词: 光伏发电; 功率; 超短期预测; 双向长短期记忆网络; 卡尔曼滤波器中图分类号: TB9; TM619 文献标志码: A 文章编号: 1674–5124(2025)05–0141–07
Abstract: The ultra short term prediction of photovoltaic power generation provides support for the dispatch of other adjustable power sources such as coal-fired power and energy storage in the power grid. A hybrid prediction method combining bi-directional long short term memory (Bi-LSTM) and Kalman filter is proposed to address the issues of low accuracy in predicting photovoltaic power generation due to the randomness of meteorological factors and the accumulation and aging of photovoltaic cell arrays. The Bi LSTM model learns the characteristics of meteorological factors and combines them with weather forecast data to reduce the random errors caused by meteorological factors. Kalman can reduce the cumulative errors caused by factors such as dust accumulation and aging in photovoltaic cell arrays. Example verification shows that under longterm operating conditions, the hybrid model improves prediction accuracy by 3.78% and 2.50% respectively compared to single Kalman and Bi-LSTM models.
Keywords: photovoltaic power generation; power; ultra short term prediction; bi-directional long short term
memory network; Kalman filter
0 引 言
精确的光伏发电功率预测为电网调度煤电、储能等其他可调电源提供支持。(剩余10961字)